Sergey Vlasov
2000-Dec-15 18:00 UTC
OpenSSH 2.3.0p1: Problem with the init script in the RPM
There is a problem in the contrib/redhat/sshd.init script: it starts sshd without specifying the full path (/usr/bin/sshd for the RPM installation). The daemon starts up and works, but dies when it receives a SIGHUP, because it is unable to re-exec itself without the full path. -------------- next part -------------- --- openssh-2.3.0p1/contrib/redhat/sshd.init.initfix Mon Oct 16 05:25:17 2000 +++ openssh-2.3.0p1/contrib/redhat/sshd.init Fri Dec 15 20:49:53 2000 @@ -57,7 +57,7 @@ echo -n "Starting sshd: " if [ ! -f $PID_FILE ] ; then - sshd + /usr/sbin/sshd RETVAL=$? if [ "$RETVAL" = "0" ] ; then success "sshd startup"
Pekka Savola
2000-Dec-15 18:09 UTC
OpenSSH 2.3.0p1: Problem with the init script in the RPM
On Fri, 15 Dec 2000, Sergey Vlasov wrote:> There is a problem in the contrib/redhat/sshd.init script: it starts > sshd without specifying the full path (/usr/bin/sshd for the RPM > installation). The daemon starts up and works, but dies when it receives > a SIGHUP, because it is unable to re-exec itself without the full path.This has been noted before (by me, at least). The problem is that sshd might not always be in /usr/sbin; for example, if you redefine %{_prefix} in your rpm macros to make all RPM's go under /usr/local (as a company policy, for example..), the script won't work. These people are a small minority, though. The change would be good for most people, but it might create problems for others. -- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords